Assessment of Landslide Vulnerability in Urban Areas Using GIS and Remote Sensing: A Study in Ambon City

ABSTRACT


INTRODUCTION
Landslides have garnered significant attention due to their status as one of the most prevalent natural disasters worldwide in terms of both human casualties and socioeconomic devastation (Nefeslioglu et al., 2008;Shahabi et al., 2014;Benchelha et al., 2020).These events primarily stem from physiographic conditions and commonly manifest during rainy seasons (Moreover, rapid population growth (Lombardo et al., 2019), has exacerbated their impact, leading to thousands of deaths and substantial infrastructure damage annually across the globe (Juang et al., 2019).Landslides not only result in fatalities and structural destruction but also possess the potential to alter landscapes significantly.Regional topography, soil composition, vegetation, and land use significantly influence and hasten the occurrence of landslides (Lavan et al., 2021).
Yearly, landslides constitute a recurring natural phenomenon worldwide.For instance, these events claim around 200 lives annually in the Himalayan, resulting in economic losses surpassing the US $1 billion (Tran et al., 2021).Meanwhile, according to the National Geological Hazard Bulletin of China, between 2007 and 2016, an average of 762 individuals were reported dead or missing each year due to intense landslides (He et al., 2020).In Indonesia, based on data from the National Disaster Management Agency (BNPB) from 2011 to 2015, there were 2,425 landslide incidents across provinces such as Central Java, West Java, East Java, West Sumatra, and East Kalimantan (BNPB, 2016).
Natural hazards refer to perilous natural occurrences within a specific time and space (Bhat et al., 2019).Among these, landslides represent hazardous events involving the movement of rock masses, debris, or soil down a slope under the influence of gravity (Varnes, 1978;Guzzetti et al., 2005;Benchelha et al., 2020).Often observed on hillsides (Ahmed et al., 2020).The initiation of slope movements results from intricate forces acting within the rock or soil mass on the slope (Cruden, 2018).wherein movement transpires when shear stress surpasses the material's strength, differing from soil erosion mechanisms (Devi, 2020).This concept of landslides encompasses the movement of material down a slope (Enigda & T, 2021).
In recent decades, advancements in remote sensing techniques and Geographic Information Systems (GIS) have significantly contributed to delineating areas prone to landslides, particularly in mountainous regions (Tewari & Misra, 2019).Additionally, GIS facilitates spatial data processing crucial for creating landslide hazard inventory and zoning maps (Van Westen, 1993;Singh, 2013;Uvaraj & Neelakantan, 2018).
Remote sensing involves capturing, measuring, and analysing images and digital representations of energy patterns emitted from sensor devices without direct contact, aiming to gather precise information about objects and the environment (Yadav et al., 2016).These systems are categorized into two groups based on technical solutions, with passive systems measuring existing radiation, such as solar radiation (Martensson, 2011).On the other hand, Geographic Information Systems (GIS) are utilized to collect, process, and integrate data and rapidly display outcomes in geographically referenced maps and reports (Sing et al., 2016).Ambon, Indonesia, characterized by its physiography, predominantly comprises hilly to mountainous terrain, encompassing approximately 89% of the area with steep slopes, while only about 11% constitutes plains.This physiographic setup often triggers landslides, a natural geomorphological process inherent to mountainous landscapes (Wang & Li, 2017).However, the limited available land in Ambon City intensifies land conversion, particularly in hilly areas that typically serve as conservation zones.
Researchers have extensively investigated landslide occurrences in various locations using diverse analytical techniques and approaches.(Tran et al., 2021) conducted mapping of landslide vulnerability employing Naïve Bayes (NB), Multilayer Perceptron (MLP), and Alternating Decision Tree (ADT).(Lavan et al., 2021) Utilized Geographic Information Systems (GIS) to explore the correlation between rainfall runoff and landslides.(Tanizaki & Ayu, 2021) utilized remote sensing and GIS alongside the Analytical Hierarchy Process (AHP).(Enigda & Suryanarayana, 2021) assessed slope instability issues using the Main Ethiopian Rift (MER), while (Gong et al., 2021) devised a method to analyze landslide stability derived from rainfall and vegetation root systems.
The novelty in this research lies in the analytical approach that integrates GIS techniques with satellite imagery, topographical information, and other geospatial data to identify and classify landslide vulnerability in Ambon City.This study doesn't solely rely on a single landslide-causing factor.Still, it integrates several factors such as rainfall, slope inclination, soil type, and rock type to understand the risk level holistically.Furthermore, the analysis of patterns and distribution of landslide-prone areas is conducted by linking these factors with the development of built-up areas over time, providing a deeper understanding of the impact of urbanization on natural disaster vulnerability.
The comprehensive integration of data and a multifactorial approach in this research strengthens the understanding of landslide vulnerability complexity, making it distinct and innovative compared to previous studies that tended to focus on only one or two factors.

RESEARCH METHODS
This research employed both qualitative and quantitative analytical methodologies with a spatial approach.The study encompassed interpretive and survey-based research methods, analysing primary and secondary data sourced from satellite imagery, on-site observations, and relevant agencies.A survey strategy was adopted, emphasizing observing and measuring variables essential for landslide analysis.The research was conducted from July to September 2022 in Ambon City (Figure 1), covering five administrative districts: Sirimau, Nusaniwe, South Leitimur, Ambon Bay, and Ambon Baguala Bay, for data collection and observation purposes.In the effort to determine the classes of land sliding, a guideline based on the level of "Harkat total" is utilized.This approach aids in categorizing data according to the intensity or level of the evaluated aspect.This procedure involves grading across five parameters, where the highest harkat amounts to 15 and the lowest amounts to 7. To establish four classes, a divisional interval of three is required.Thus, the classes of land sliding can be determined using a four-unit value interval, as depicted in Table 3.
The utilization of this guideline enables researchers or stakeholders to structure data more systematically and classify the intensity of land sliding in greater detail.By establishing class ranges based on the predetermined Harkat total parameter, this method facilitates a more accurate assessment of landslide risk in https://doi.org/10.24114/jg.v16i1.41978Lasaiba, M.A et al. (2024) Assessment of Landslide Vulnerability | 36 specific areas.These steps aid in depicting land conditions more clearly and support mitigation efforts or preventive measures that can be implemented to reduce the risk of land sliding impacts in a particular region.Field observations were conducted during the implementation phase to validate the accuracy of image interpretation, ensuring alignment with the actual field conditions, and measuring parameters that couldn't be ascertained from the images.
Surveys were conducted across the research area, particularly focusing on regions where land use density or weather factors, such as cloud cover, impeded precise image interpretation.
Validation was performed to assess interpretation accuracy by comparing the interpreted results from images with fieldchecked results.Activities were undertaken to enhance interpretation outcomes and mapping accuracy through sampling.Landsat satellite image data processing involved layer stacking, consolidating eleven different channels into a single dataset for easy analysis and comprehensive interpretation.
Radiometric correction procedures were implemented to minimize errors resulting from the recording system and the passage of sunlight through objects to the recording camera.Radiometric accuracy denotes a system's capability to discern differences in electromagnetic energy, relying on the detector's signal-tonoise ratio and its capacity to convert continuous electromagnetic signals into digital ones.After radiometric correction, the image may undergo geometric correction to remove spatial and geometric distortions, which involves mapping the image to geographic coordinates or an appropriate projection.Equations and steps for geometric correction will vary depending on the method used.
The data gathered in this final stage is deemed suitable for analysis and serves as crucial    The built-up area in 2012 4,527,424 2.
The built-up area in 2022 5,707,990 Source: Data Processing (2023)

Landslide Factor Analysis Rainfall Factor
Rainfall plays a significant role in triggering landslides, particularly in regions like Indonesia characterized by a wet tropical climate where it stands as a primary determinant of the climate.It serves as an external factor outside the slope's body that can lead to landslides due to its intensity and subsequent flow in various locations (Handoko & Ikaputra, 2019) The assessment and calculation of rainfall intensity values are presented in Table 5. Figure 2 illustrates the spatial distribution of rainfall across the study area.Rainfall, a universal natural occurrence essential to various aspects of life, holds significant influence in Indonesia, especially in regions like Ambon City characterized by a tropical wet climate and frequent heavy rainfall.Intense rainfall often leads to increased surface water flow, resulting in substantial soil erosion and the potential displacement of soil material, subsequently compromising slope stability.According to (Arsyad et al., 2018), the precipitation amount on a slope tends to rise with altitude, rendering slopes devoid of vegetation or impermeable layers highly susceptible to landslides during heavy rains (Rienzi et al. 2013).The magnitude of rainfall directly impacts factors such as soil distribution strength, its carrying capacity, and vulnerability to damage (Hutapea, 2020).Studies by (Andriawan and Sarya, 2014) indicated that rainfall intensities exceeding 50 mm/h often trigger shallow landslides, while research by (Hidayat and Zahro, 2018) identified rainfall data as a catalyst for landslides in the Banjarnegara Region, particularly emphasizing that daily maximum rainfall of 56 mm could induce landslides.Furthermore, (Gemilang et al., 2017) noted that areas like the Bungus Hills, experiencing an average rainfall of over 200 mm, exhibit a notably high level of landslide hazard.

Slope Factor
Slope inclination is a critical factor contributing to landslides.It represents the ground surface's stability against gravitational forces (Fransiska et al., 2017).The determination and categorization of slope values are detailed in Table 6.6 illustrates that Ambon City exhibits diverse slopes, with the majority of the area comprising slopes very tilted than 30% and ranging between 15-30%, encompassing approximately 10,750.05hectares.The terrain conditions in Ambon City, primarily consisting of slopes categorized as values 3 and 4 or with high percentage slopes (as depicted in Figure 2), pose a significant risk for potential landslides.
Slope inclination stands as a pivotal element influencing the occurrence of landslides, observed consistently across diverse global regions, including Indonesia, where the instability of steep or excessively steep slopes often leads to landslides (Fransiska et al., 2017).(Rompon and Almulqu, 2018) underscored the tendency for landslides to manifest more frequently in areas with elevated slope gradients.The slope factor contributes to diminishing the soil's shear strength, rendering it susceptible to collapse, as highlighted by (Akbar et al., 2022).Moreover, the inclination of the slope directly impacts the magnitude of landslides, evidenced by (Çellek, 2020) demonstrating an escalation in soil mass movement corresponding to an increase in slope, attributable to heightened gravitational thrust and shear stresses.(Nengsih, 2015) reinforced this notion, indicating that slope stability hinges upon the interplay between soil shear strength and shear stress, where soil collapse ensues when shear stress surpasses the soil's inherent strength.

Land Use Factor
The factor of land use encompasses the various human activities and natural elements covering the soil surface, such as vegetation and rock structures.As indicated by (Nugroho et al., 2017), different types of land use significantly influence the stability of slopes.Land use constitutes an external trigger that impacts the slope.Ambon City spans an area of 32,068,753 hectares, encompassing five primary land use types: forests, mixed gardens, shrublands, built-up areas, and plantations.Table 7 presents the classification of land use values in the area.2), significantly influence landslide occurrence frequencies.
Alterations in land use transitioning from natural conditions to agricultural, residential, or industrial purposes can bring about changes in soil and vegetation characteristics, leading to the uprooting of soil-bound roots, amplified erosion, and compromised slope stability.Poor land-use decisions, not in harmony with environmental requisites, can heighten the likelihood of landslides (Nugroho et al. 2017).According to Ritung et al (2007), regions characterized by steep slopes and specific land-use patterns, like moors and scrubs, often witness landslide occurrences.The potential degradation of slope stability contributes to increased landslides as landuse intensity escalates (Hasibuan and Rahayu 2017;Soewandita (2018).Mixed gardens are identified as high-risk areas for landslides, necessitating improved land management practices aligned with land conservation regulations.Suwarsito et al (2020) propose the strategic placement of perennial plants possessing deep root systems in sloping areas to mitigate the incidence of landslides.

Soil Type Factor
The soil type factor plays a crucial role in the occurrence of landslides.In Ambon City, the soil types are categorized into four units: 1) alluvial, cambisol, regosol, gleysol, 2) cambisol, latosol, regosol, 3) latosol, cambisol, and 4) rensina, cambisol, litosol.Unpatti (1985;Lasaiba, 2012) Table 8 illustrates that the soil types prevalent in Ambon City predominantly comprise cambisol, latosol, and regosol soil units, collectively occupying an area of 23,599,715 hectares.However, the latosol and cambisol soil units only represent a small proportion, accounting for 6.12% of the total area or 1,699,064 hectares.Latosol soil type, in particular, while covering a relatively small land area, exhibits a widespread presence throughout Ambon City.The spatial distribution of this soil type is depicted in Figure 2.
Landslides tend to happen in specific soil types, especially following rainfall.The occurrence of landslides is notably influenced by the fine and smooth texture of the soil, particularly clay-based textures.As highlighted by (Harjadi & Paimin, 2013).soil textures classified as finer are more susceptible to shrinkage, instability, or movement.(Heradian and Arman, 2015), point out that clayey soils with high water content represent areas prone to landslides due to their lower resistivity values.(Soewandita, 2018) notes that thick soil layers with a porous structure, particularly found in sloped areas, exhibit high vulnerability to landslides.(Hadiyanto, 2011) highlighted the high sensitivity of cytosol and regosol soil types to water.Conversely, soil types such as alluvial, gleysol, planosol, laterite, and hydromorphone are less sensitive to water, resulting in a lower occurrence of landslides during the rainy season.

Rock Type Factor
Rock types in Ambon City encompass various categories such as sandstone, serpentine, diabase, gabbro groups, andesite groups, breccias, loose materials, granite units, limestone units, and alluvial deposits.Table 9 presents the classification and values assigned to these rock types in the area.The rock types in Ambon City exhibit diverse distributions, with loose material covering the most extensive area, spanning 10,960 hectares or 45.05% of the area.Additionally, the Andesite, Dacite, and Breccia groups collectively cover an area of 5,684.05hectares.Conversely, sandstone represents the smallest area, covering only 1524.21 hectares or 4.73% of the total area.The prevalent distribution of loose material and the Andesite, Dacite, and Breccia groups signifies their significance as the most extensive rock types in Ambon City, attributed to the Ambon volcanic deposits during the Pliocene era.
Ambon City exhibits geological structures primarily characterized by down (normal) faults and joint faults.The fault structures, evolving from northeast to southwest directions, intersect granite rock units, and clusters of serpentine, diabase, and gabbro units situated in the headlands of Seri Village and Hukurila Village.These findings align with Rahman (2010) suggesting that locations prone to landslides are associated with rock domes exposed to flow, and soil structures consisting of older Andesite and Andesite Breccia formations affected by numerous faults.Such rocks are prone to weathering into soil, rendering them susceptible to landslides when present on landslide-prone slopes (Putra et al. 2019).Volcanic sedimentary rocks and sedimentary rocks with sand-sized grains, along with compositions of gravel, sand, and clay, exhibit weaknesses.When subjected to weathering processes, these rocks swiftly transform into soil, posing vulnerability to landslides, particularly when situated on steep slopes (Darmawan et al. 2021).

Landslide Vulnerability Analysis
Regarding landslide vulnerability analysis, the determination of landslide hazard zoning in Ambon City was conducted by categorizing it into five risk classes: shallow, low, medium, high, and very high.This zoning was established based on the landslide hazard analysis design.By summing up (scoring) the factors present in each field unit, the level of vulnerability or likelihood of landslides can be calculated.Table 7 below illustrates five interval classes composed of the variables and their respective weights used in the analysis.7 delineates that a mere 2,641,019 hectares or 8.21% of Ambon City's total area is categorized as having a very low vulnerability to landslides, representing a minimal risk level (8%).Conversely, a significant high, totaling 16,619.011hectares or 51.63% of the area, falls into the high vulnerability zone for landslides.Moreover, the middle range of the landslide vulnerability map demonstrates a vulnerability level ranging from medium to high, predominantly observed in the hilly and mountainous regions of Ambon City, especially in areas characterized by steep slopes.The spatial distribution of landslide vulnerability can be observed in Figure 3.
Moreover, the central zone depicted on the landslide vulnerability map tended to exhibit a vulnerability level ranging from moderate to high, particularly prevalent in the hilly and mountainous terrains of Ambon City, followed by regions characterized by steep slopes.This study draws parallels from research findings conducted in the Ponorogo Regency by (Yuniarta et al., 2015& Naryanto et al., 2019), an area recognized for its predisposition to landslide occurrences owing to its predominant hill-based morphological features.A similar investigation conducted by (Fitrianingrum & Ruslanjari, 2012) in the Kulonprogo Regency, specifically the Menoreh Hills area, also highlighted geomorphological vulnerability to landslides, primarily attributed to highintensity and rapid rainfall.Figure 4 depicts the precise locations of landslides within the study area, causing substantial damage to several residential buildings.The Regional Disaster Management Agency (BPBD) in Ambon City reported that a total of 17 landslides resulted in damage to approximately 56 houses.These landslides occurred across four sub-districts: Teluk Ambon, Nusaniwe, Sirimau, and Baguala.In Nusaniwe District, landslides were observed in three locations: Kudamati, Benteng, and Amahusu.Sirimau District experienced landslides in Batu Gajah, Amantelu, Batu Meja, Waihoka, and Soya areas.In Teluk Ambon District, landslides took place in the Tawiri, Poka, and Tihu areas.The Baguala sub-districts affected by landslides include Negeri Lama, Lateri, Halong, and Passo.

Analysis of Built-up Land in Landslide Disaster Area
The study utilized the development data of built-up areas and conducted an analysis of landslide vulnerability in Ambon City.This information served as input for the analysis aiming to identify built-up areas situated within regions prone to landslides.
The distribution of built-up areas within each class of landslide vulnerability was determined through an overlay process, specifically by overlaying the outcomes of the two analyses and conducting an intersect analysis.This method, known as Kawasan zoning, is detailed in the subsequent table.The distribution and extent of built-up land within each class of landslide vulnerability in Ambon City are outlined in Figure 4 and detailed in Table 8.The largest areas among the zones are Z-1 and Z-4, covering 2,224,549 hectares (39.01%) and 2,000,913 hectares (35.09%), respectively, signifying the most extensive zoning classifications.Zones Z-2 and Z-3 encompass 678,094 hectares (11.89%) and 777,107 hectares (13.63%), respectively.On the other hand, Z-5 represents the smallest area, covering only 21,691 hectares (0.38%).
Zones Z-4 and Z-5 exhibit a high to very high vulnerability to landslides due to steep slopes ranging from 25% to greater than 40%, coupled with rock types prone to weathering and the prevalence of built-up land, which

CONCLUSION
The research area's land units are categorized based on three primary factors: the steepness of the terrain, land use, and specific geological features.Landslides in this region were triggered by heavy rainfall, averaging between 27.7 to 34.8 meters, extensive steep to very steep slopes (constituting 62.53% of the area), predominant land use of mixed gardens and built-up areas (occupying 52.05% of the total area), and soil types like cambisol, latosol, regosol, alongside loose material rocks, covering approximately 73.31% and 45.01%, respectively.
The vulnerability of landslides in this area is notably high, encompassing approximately 51.63% of the total area.This high vulnerability is mainly concentrated in regions characterized by steep to very steep slopes in hilly terrains.An analysis of the built-up areas in regions susceptible to landslides, particularly zones Z-4 and Z-5, highlights slopes ranging from 25 to over 40%.These areas possess rock types highly susceptible to weathering, alongside land

Figure 1 .
Figure 1.Research Location a. Ambon Island Map, b.Administrative Map Ambon City and c.Province Map Maluku (Source: Data Processing, 2023).
material for comprehensive examination.The overall data compilation involves grouping and systematically analyzing data in a quantitative and deductive manner.Spatial analysis techniques were employed to discern spatial patterns within the collected datasets.The mapping process for each indicator was derived from the 2004 Puslittanak estimation model.Puslittanak, a research institution operating under the Indonesian Agency for Agricultural Research and Development, specializes in studying soil resilience and climatology in Indonesian agriculture.Utilizing this model, parameters are categorized based on their respective scores, which are then aggregated to ascertain geographical suitability.This process results in the assignment of five classes depicting landslide vulnerability levels: very low, low, moderate, high, and very high.

Figure 4 .
Figure 4. Rainfall Intensity Map, b) Slope Map, c) Land Use Map, d) Soil Type Maps, and e) Map of Rock Types

Figure 5 .
Figure 5. Map of Landslide Locations in Ambon City areas possess low soil retention capacity, rendering them highly susceptible to erosion.According to the Regulation of the Minister of Public Works No. 22 of 2007, constructing settlements is only recommended on slopes ranging from 0 to 15% (flat to slightly steep), designating zones Z-1, Z-2, and Z-3 as suitable and safe for built-up land use.Development in areas like Z-4 and Z-5 with steep slopes requires specific criteria, including engineering measures like embankments to maintain slope stability.However, the extreme impacts of climate change could affect the resilience of these engineering interventions, possibly necessitating increased costs for handling such construction projects.Most built-up areas within zones Z-4 and Z-5 have been established by communities highly susceptible to landslides, making relocation impractical.In the future, the government must proactively control the expansion of built-up land in these high-risk zones, focusing on policies and spatial planning.Development efforts should be directed towards areas with low to moderate landslide vulnerability, such as Z-1, Z-2, and Z-3 in Ambon City.Moreover, public education about the consequences of development activities leading to landslides is crucial, especially in high and very high vulnerability areas.

Figure 4 .
Figure 4. Map of Landslide Prone Areas of Land Zoning in Built-up Areas in Ambon City Source: Data Processing (2023)

Table 4 .
Area of Ambon City Built-in 2012 and 2021

Table 6 .
Class and Slope Area

Table 8 .
Class and Area of Soil Types

Table 9 .
Broad Class of Rock Types

Table 8 .
Area of Built-up Land in Landslide Disaster Area